Appears in collection : 2022 - T1 - WS3 - Mathematical models in ecology and evolution

HIV-1 is treated with combination therapies of multiple simultaneous drugs targeting different stages of the viral lifecycle, such that no single mutation confers resistance to all drugs used in a treatment. Complete drug resistance should require the co-occurrence of multiple resistance mutations in a single reverse transcription step - an extremely rare event probabilistically. Nevertheless, HIV does evolve resistance even on combination therapy, suggesting an incomplete understanding of the dynamics at play. In this presentation, I'll describe several key features identified in clinical trial and genetic data that a model of drug resistance evolution must match - namely resistance mutations emerging one at a time in a partially predictable order, even years after therapy onset. I will then describe our efforts to model this process of multi-drug resistance emergence, using temporally- or spatially-varying drug levels within individuals. We find that a model of spatial heterogeneity more straightforwardly matches the patterns found in clinical data, and I close by discussing how we can use these findings to design more evolution-proof combination therapies.